Neuro-robust controller for non-linear systems

This work shows a new method to obtain the tuning parameters of a controller based on PID topology in standard format in order to lead the controlled system to a stable state. This technique is based on the well known Gain Scheduling method. Firstly, the system is identified for each key operating point obtaining the corresponding transfer functions by means of least squares techniques. Then the stability structures are calculated using the transfer functions aiming to train an artificial neural network which prevents the system from becoming unstable. Finally, the proposed method is experimentally verified on a laboratory plant.